Practice English Speaking&Listening with: Virtual Peering Series – Central Asia #2

Normal
(0)
Difficulty: 0

Hi everyone, ladies and gentlemen, it's a real pleasure to welcome you to the second

event in the Virtual Peering Series for Central Asia. And you're perfectly aware

of the fact that the entire series was created by a bunch of people that I will

introduce to you shortly. I am the External Relations Officer for Central

Asia and Caucasus, RIPE NCC, and this series was organized jointly by ISOC, RIPE

NCC and the Euro-IX to facilitate development of peering, and establishment

of better relations amongst Central Asian states. This series will help our peers in

Central Asia to deploy peering networks and IXPs, and we've been assisted by

FLEXOPTICS, AMS-IX, Facebook, PCH, and, well, we're get more details on our

sponsors later in the session. But I would like to ask everyone to be as active as

possible, there will be a third event in the same series, which is slated for

September. And the third event will probably be devoted to IXP regulations,

and the legal framework. But today, there will be another great presentation to be

delivered by PCH. And you will also have a chance to discuss a number of issues to be

raised in the PCH's presentation for you to be able to better understand what needs

to be done in Central Asia. Without further ado, let's move on according to

the agenda,

Hi everybody. My name is Nishal Goburdhan. I'm an Internet analyst and I work at

Packet Clearing House, we're a nonprofit that's based in California, and we've been

helping Internet exchange points develop across the world since about 1993. We're

also involved in some DNS work. But this presentation is not about PCH, if you'd

like to know what we do, you can contact me at the end, and I'll happily talk more

about that.

I think this is a picture that we're all familiar with. It's source is caida.org,

and it largely tries to describe what the Internet looks like. But practically, I

don't know about you, if I look at this picture, it doesn't tell me a lot about

the Internet, except that there's lots of lines interconnecting people. When we

think about the Internet, as normal humans that are interacting with it, is usually

something like this, where there's a certain tier of Internet networks that are

connecting to each other. And those Internet networks usually work in a way

that they have to perform, or buy transit from one another.

At its most basic level, the Internet is, as you know, probably typically a local

Internet service provider, somebody that sells Internet access in your region. That

Internet Service Provider typically connects to a larger Internet service

provider, perhaps one that has got operations in multiple cities, or in

multiple countries, but it's still typically limited to a region. That second

tier service provider typically ends up buying IP transit from somebody that we

call a first tier service provider, or a larger Internet service provider, a global

Internet service provider. There are about 14 of these in the world today. The thing

that's common about this really is one thing, that in order to get access to the

Internet, if you're somewhere at the bottom of the chain, you've got to buy IP

transit.

At the end of every month, at the end of every quarter, the end of every year,

you've got to go to that Internet service provider that you have a contract with,

you've got to give them money. And in exchange for you giving them money, sorry,

they will connect you to the Internet. It's the model we were taught. It's the

model that we grew up with. And it's the model that we know that works. And it's

really the simplest way that we know of running the Internet.

What started happening about 25 years ago was that the entire network started to

realize that they could try to potentially save money, by not just passing on this

kind of IP transit to networks that were higher up in the food chain. What if they

started interconnecting to each other? And that's really what those purple lines are.

Those purple lines are interconnections between the Internet service providers

themselves at a lower tier, starting to understand that it was more efficient, and

faster and cheaper. more importantly, if they could interconnect between

themselves. And this has happened over time, and this ultimately is the Internet

as we know it, where you've got a neutral interconnection point, and I've depicted

that as the white cloud in the middle. You've got the smaller tier ISPs all

interconnecting into that, and they're peering by means of those purple clouds.

Now I have been using two words consistently, and I haven't yet defined

them. And I always like to have a clear definition of what we're talking about,

because I think that that's going to help all of us understand what we're trying to

get to. And those two terms really are Internet transit and Internet peering. The

simple concept that I'd like you to remember is that Internet transit is a

relationship where you have a commercial commitment to pay somebody in exchange for

them, giving you access to as many parts of the Internet as they can get you to.

Peering agreements are typically agreements that happen across usually

Internet exchanges. And the fundamental difference that we want to bear in mind,

peering agreements usually are settlement free. And that's a fancy way of saying

that there's no money that goes across a peering agreement. So, you can read the

full on definition later, obviously, the slides will be made available to you. But

the essential difference that we care about is that transit costs money, and

peering generally doesn't. It's not true that peering is free, let me just clear

that up at the start, you obviously need to invest in your infrastructure. But

we're talking about the nature of the relationship between the two parties,

right?

It's a little known secret in the Internet service provider world that there's really

only three ways for traffic to flow between Internet networks. And that really

is that if you're a big Internet service provider, and you have a customer in your

network that's trying to talk to another customer in your network, that's the best

possible situation for you to have. Because if you're the big green ISP that

I've pictured here, and you've got a customer on the left that's trying to

connect your customer on the right, the customer on the left pays you, let's just

say they pay you $1, the customer on the right pays you, and they also pay you a

$1. At the end of the day, you've got to make the network between these two

customers work, which is quite easy, really, because it's your network. And, at

the end of the day, you earn $2. And that's fantastic for you. So situation,

one, you bought your customer trying to speak to your customer.

But no single Internet service provider is big enough to connect the entire world

together. And as you've seen in the picture earlier on, today there's about

77,000 Internet networks that make up the Internet. So, ultimately, for you to get

to speak to another network, you're going to have to connect to somebody else. And

the model that we know about is, we go to a transit network, and that transit

network will connect us to the Internet. So, here the picture's changed slightly,

we've got our green customer on the left, that hasn't changed. And we've got a

customer on the right, the red customer, and we've got two networks inbetween, with

the transit network, or even many transit networks, it's really not important. What

we see here is that both the red and the green networks earn money from their

customers, because this is how they are staying in business. This is how they

connect their customers to the Internet. But because there's no direct path between

the green and the red Internet service provider, what they have to do is they

have to go through a transit network. And as a feature of going through the transit

network, they have to pay out money. At the end of the day, they recognize revenue

from their customers, but they've got to pay the transit network. And the important

part here is that the earnings that they get is that $1 that they got from the

transit customers, less whatever they have to pay to the transit networks.,

So, situation two, you earn 1 minus x, which is less than two. The third possible

situation we have, is where we have the same two networks, but they are both

connected to the same Internet exchange, or multiple Internet exchanges, that

really doesn't matter to us. And the notable difference here is that because

they are peering across the Internet exchange, they're not paying each other

money to exchange the traffic. In this case, each of the green and the red

networks still earns the $1 that they earn from their customers, but because they're

not really paying money, there's no commercial agreement between the red and

the green network, they retain that dollar. Of course, they're spending some

on their network, but the fundamental differenc is that they still earn $1.

So, you've got three situations. One where you earn $2, one where you earn 1 minus x,

which is whatever that's going to be, and where you earn $1. Any rational network

will always try to make the two, right? Let's be honest about this, people want to

earn as much as they can. But because it's not practical, it's not possible for one

Internet provider to connect to the entire world, what's realistically going to

happen is that the rational networks are always going to seek to maximize their

earnings. And in a situation where you can't earn $2, the most rational thing for

you to do is to try to earn the next best thing, i.e. earn $1 and you will always

try to maximize your peering.

If that's not something that you're doing now, as a network operator, it's something

that you will start doing because, as I pointed out, there's really only three

ways for traffic to flow across any network. We do this by means of what we

call an Internet exchange point, and you're going to hear a lot more people

discuss Internet exchange points.

The only thing that I really wanted to make clear is that an Internet exchange

point is physical infrastructure. In the modern terms, we have either gigantic

ethernet switches, or a series of ethernet switches, at Internet exchange points. And

those ethernet switches, or series of switches, bring together a whole lot of

value. Some of the benefits that you see, and the panelists are going to talk a

little bit more about them, I've listed here. I won't go through these because as

I said, they're going to be discussed a little later in the presentation.

Where these Internet exchanges are across the world? This is a map that I pulled up

this morning from the Internet exchange point directly that PCH operate. We've

been doing this for, I forget, 25 plus years now. You can see that there's a

really dense set of Internet exchanges in Western Europe, in the United States, and

in Brazil. The rest of the world, we're still catching up, not necessarily a bad

thing. Because if you take this map, and you map it on to population sectors,

you'll find an Internet exchange point pretty much where there's high density

population. And that's something to factor and consider when we talk about where and

how we build Internet exchanges, which is going to be in the last part of this

presentation.

So I'm going to pause here for a bit and I'm going to let the panelists weigh in on

the first very important question, which is why do we peer?

Indeed, why do we peer, or why do we need peering in the first point? First and

foremost, peering allows you to drive down your expenses and reduce your costs. It

makes economic sense. It is linked with national security issues. It also ensures

the right performance and it develops the economy. But anyway, why do we peer? The

question that we would like to pose to you, and discuss with you. Any views on

that? What's your take on why we need peering. Talant, how about yourself?

Right. Honored to be the first one to take the floor. My name is Talant. And I'm

based in Bishkek, Kyrgyzstan. I'm chair of the chapter of ISOC. And Aziz, a colleague

of mine, is also online.

Actually, this is a question for the entire audience for the participants. Why

do we peer? Why do we need to peer in Central Asia? Many people say that we have

already established some links and direct connections. Why do we need to peer with

everyone, or why do we need peering at all? What's your take on that? Are there

any raised virtual hands?

No, I've checked I haven't seen one. In any case.

Right. In that case, I guess we need to deep dive into this issue, since we don't

know what the answer to the question is, if no one knows.

Yeah, Bakhrom says peering is required to bring down Internet connectivity costs and

to reduce the delays. Bakhrom, would you ike to take the floor and expand and foll

w up on that? Bakhrom is based in Taji istan, and he is an active part

cipant in all of our events. He is also an active participant in various indu

trial events.

Hi everyone, Bakhrom Nasirjanov is my name. Thank you for introducing me.

Indeed, I'm based in Tajikistan, and I work for mMegafon Tajikistan, which is a

carrier in Tajikistan. As far as peering is concerned, based on my experience, I

can tell you that peering is quite useful, and it becomes part and parcel of Internet

connectivity between various ISP is in the same country, or among ISPs based in

adjacent countries. Because those countries that could be considered to be

at the edge, if they do have transit Internet traffic, for them Internet

connectivity is too expensive. Therefore, if you don't have any peering agreements

with other ISPs based in other countries, for the edge-based countries, it's

difficult to transport traffic from point A in one country to point B in the same

country, they have to go through European countries or Russia. I'm talking about

Central Asian ISPs primarily.

Naturally that increases the Internet traffic costs, since they have to go to

the backbone networks. In addition to that peering reduces in other ways. We border

on China. But unfortunately, we have not signed a peering agreement with China.

Therefore, to reach Chinese resources, we have to transit our traffic through Russia

and via other Central Asian countries, sometimes we have to go via the US. But if

we had a direct connection with China, we could reduce the the delay down to 50

milliseconds or even lower. Right now it's 400 to 500 milliseconds. And this is one

of the huge problems that could be solved with the help of peering.

Thank you Bakhrom. Talant started saying something but I guess he ran into some

technical glitches. Talant, now the floor is all yours. What's your take on that?

Thank you so much, Vahan. I just wanted to briefly go over our experience.

A couple of years ago, we started building the IXP in the Fergana Valley. When we

During the past few years. The Internet costs went down. The quality of the

started building the IXP in Kyrgyzstan in the Fergana Valley, the Internet

connectivity in our country was one of the most expensive ones across the world, I

would say. And, to reduce the Internet connectivity costs, especially for the

Internet connectivity went up. And, thanks to RIPE NCC and ISOC, we organized a

nationals living in the Ferganar Valley, we decided to deploy an IXP in the city of

Osh. Kyrgyzstan is a specific country, because we are split geographically wise

into the northern and the southern parts. We had an IXP in Bishkek, which is in the

northern part. It has existed in the country for almost two decades. It was the

only IXP in the country. And when we started deploying the second IXP, more

IXPs started coming on stream. Right now there are quite a few of them. Most of

them are private, I would not call them independent, but they are private, which

is still good for developing the Internet market in the country.

number of educational events, and we tapped on the yet untapped potential. This

is it.

Thank you, Talant. Aziz, I believe you wanted to say something. Aziz Soltobaev,

you the floor.

Hi everyone. I'm gonna speak Russian. I also work for the Kyrgyz chapter of ISOC,

and I'm responsible for developing IXPs in Kyrgyzstan. The digital sustainability is

improved by IXPSs. IXPs also increase the information security, especially if IXPs

are combined with a localizing international service such as DNS root

servers.

And in Uzbekistan, we try to deploy SNS-IX IXP, the main objective of which is to

Thank you, Aziz. We have Talgat online. Talgat from Kazakhstan also wanted to

improve the interconnectivity. A couple of root servers will come on stream shortly,

and we really hope that in the near future, jointly with PCH, we will localize

share a couple of comments. The floor is all yours.

a couple of other services. We will team up with RIPE and Kira. Ultimately, it will

increase the sustainability of Uzbekistan and Central Asia in general. In a

Hi, everyone. I work for ISOC Kazakhstan. At the previous Virtual Series event,

nutshell, this is it, interconnectivity, and resilience are the two buzzwords and I

guess these two boss words can describe perfectly well I experienced in the

region. Thank you so much.

someone from Russia mentioned that Kazakhstan officially bans IXPs. This is

not really so. I haven't been able to clarify this issue yet. I went to the

Ministry of Digital Development, but I'm still waiting for their answer. You were

mentioning proper things, I would say. You are that an IXPs improve interconnectivity

thus improving the sustainability of the Internet, they streamline and optimize the

routing, and so on and so forth.

However, given the historic specifics, Kazakhstan is a hub for Central Asia, as

you're well aware. Therefore, the entire traffic in Central Asia is routed through

Kazakhstan. This comment is addressed to you primarily. I asked RIPE repeatedly to

do a study on the traffic routing in Central Asia. Let's say, if the traffic

goes from Tajikistan or Kyrgyzstan to Kazakhstan, does it go through Russia or

Europe? Such a study would be quite useful. I guess everyone would be

interested in seeing the outcomes. Kazakhstan is a natural hub in Central

Asia. Although I guess not everyone is happy with the work done by the Kazakh

Internet service providers, and they are not happy with the prices. This is

especially true for our Kyrgyz counterparts. However, as far as the

Central Asian IXP is concerned, I believe that it should be deployed in Kazakhstan.

It's a natural decision. And the Central Asian cooperation is something that we

should focus on. It would help us greatly.

Thank you so much, Talgat, for raising a very important issue, which is the legal

framework and the legal regulation. So, we will definitely discuss it in greater

detail at the follow up event. Regulations are important, especially in Central Asia.

I'm going to give the floor to Stavros now, and Stavros will probably share his

views on that. Stavros, you have the floor.

Thank you very much for the invitation to this very interesting panel discussion.

Among the benefits that have been already discussed and provided by the previous

participants and panelists, I want to raise also the fact that IXPS, and

deploying IXPs to localize traffic, also increases the robustness of the Internet.

In the current Internet ecosystem, which is a little bit of madness outside, with a

lot of DDoS attacks happening every minute around the globe, we can see clearly from

our customers that they suffer a lot from the transit links, that they get high

volumes of DDoS attacks that actually can saturate the transit links, means that

they also have to sometimes take extreme measures to shut down the transit links in

order to mitigate these DDoS attacks. However, if you have some peering with

local ISPs, or if you peer with them via IXPs that can help you to keep some

traffic, and still route traffic between customers and other members. That means

that the Internet becomes more robust. So, in case you get a DDoS attack from China

or Brazil, or wherever, you can shut down your transit link and still you can

communicate with other ISPs, or local banks, or other institutions, and then you

can still offer some of your services, and these people will not realize anything.

And we can see this also in the Europe, whereh IXPs are very important for local

ISPs. They can keep the traffic local, and still offer services to end users and

customers, and also in US and other places.

I would like, also, to highlight the benefits in the local communities when

local ISPs do peering, and do peering via IXPs. Back 15 years, when we deployed

AMS-IX in Curacao, the local community was suffering from very low quality of the

Internet because all the traffic even between the local ISPs was going via

Miami, and thatwas increasing a lot of the latency and the delay. The things that

happened, and improved the quality of the Internet for the local community in

Curacao, and that's also an example, of course, is that AMS-IX went to Curacao and

deployed the local IXP instance over there. Then we invited all the local

Internet providers and small network providers to start peering via AMS-IX and

they started doing so. Then the traffic localized, and the latency started

minimizing, and the local community started having a much higher quality of

Internet. Also, we deployed Google Global Cache over there, so some caching was

happening from content providers, and also increased the quality of the Internet for

some services. Some studies were executed later on, and they saw how much important

was those two moves to increase the quality of the Internet for the local

community.

But that was just an example, of course, I'm pretty sure similar things will happen

in the future, when we are going to deploy more IXPs in Central Asia, and the local

communities will benefit from more robust Internet and more faster services. That's

from from myself,

Nishal, the floor is yours.

Thank you. So it's certain, really, that this is the Internet lifecycle from an

Internet service provider's perspective, and that's usually because I speak mostly

to Internet service providers, but it's really true for any network operator.

What I'd like to do is walk very quickly through how the life cycle looks like when

you're not peering. This is the model that we know, we know that there is an Internet

service provider network, and that Internet service provider network has

customers. And to get connected to the Internet, that Internet service provider

network is going to buy transit, probably from a single Internet transit provider,

and usually from an Internet transit provider that is out of your country.

You're probably going to go to one of the big service providers, that's from one of

the more industrialized countries. That's how we all started, right, this is how the

Internet pretty much got to everywhere that we know.

What you need to see in this picture is that you're acting less and less as an

Internet service provider, and you're acting more and more as a currency, or

foreign exchange point, because what you're doing as an Internet service

provider, is you have a signed contract, in dollars or euros or rubles or whatever

it is, with an external party. You have a lot of signed contracts with customers

that are inside your environment, customers that are inside your country.

Those contracts are usually in your local currency, whatever your local currency is.

And all that you're doing is you're aggregating customers, you're aggregating

a lot of many contracts that you've signed, usually on a short term basis.

yYou're converting local currency into some kind of foreign exchange service that

you have to provide to some external provider.

I live in South Africa, and I can tell you that, over the past 10 years, our currency

has depreciated against pretty much any large major currency significantly. So, if

you live in a developing world country, you probably know the state of your

currency against more developed currencies, you probably have a really

good idea what's happened to the currency over a longer period of time. So, you know

that continuing to pay a foreign agent, with money that you are raising in your

domestic context, is not a long term solution to your problem.

Now, I realize that most of you here are probably network operators, and you're

thinking - What is wrong with this guy? We came here to listen to talk about peering

and Internet exchange points. But, for me, this is a fundamentally important thing.

That, aside from all of the benefits, you've heard about peering and localizing

content, and all the benefits you're going to hear from even the folks that haven't

spoken yet, the one thing that we tend to forget very easily is that it is an

extremely important mechanism for you to retain domestic currency.That's important,

particularly if you're in a developing country.

The lifecycle of an Internet network is usually that you migrate from one network

operator to two redundant transit providers. You do two things when this

happens. The first is, obviously you get redundancy, you have the opportunity to

say if circuit one to provider 1 fails, I have a separate link to circuit provider

2, and that's going to continue to give me uninterrupted service to the Internet.

This is really why we buy redundancy. But, whether you realize it or not, you do

another very important thing when you purchase from a redundant transit

provider. What you do is you shorten the cost, or you shorten the path, that you

use to get to a resource on the Internet.

So, here's a very simple explanation. Let's imagine you're the network at the

bottom, the network that I've labeled You, and you've got one transit network. This

is a very simple situation where you've got one path to the Internet. You're

trying to get to a popular resource on the Internet, maybe a search engine, and

that's network A, and maybe you're trying to get to a social networking site, and

that's network B. The numbers that I've chosen at random, are just the number of

hops, or the number of networks, it's the path that you take to get to those

networks. 5 and 7 is 12. Because you're connecting to two sites, the average

length for your users to get to the Internet is 6 in this case. I realize this

is a very simple example, but it's going to make a point, I promise you.

So, two important resources that your users are trying to get to, the average

length of - the number of networks, the path that your users have, is 6. When you

add a second network, a second transit network, that transit network is probably

buying transit from a different set of networks, it's probably peering at a

different point on the Internet exchange, and so it's going to have different paths,

the different costs, to get to the same set of resources. Your network now has two

ways to connect to the Internet, to those resources that we were talking about,

networks A and networks B. All things equal, the path that your network is

usually going to take is going to be the shortest path, the cheapest path, the one

that's, in the context of our example, the one with the fewest number of hops.

That means you're going to pick a path that I've highlighted here. The average

length of you connecting to the network is going to decrease, and that's now going to

be four hops away. Again, this is a simple example but it shows you one really simple

thing about the Internet. One fundamental thing that matters to your networks, users

automatically because of the technologies used in the network, right, because of the

technologies used in BGP that you're familiar with, that we're not going to

talk about here, because of technologies used by the networks at the end, that tend

to want to prefer lower latency paths. Users are going to prefer the path of the

network that is cheaper, that is shorter, they're always going to choose a shorter

path to get to you.

That's an important concept. Because when you get to the third part of the Internet

lifecycle, the part that I like to say where you add a single Internet exchange

point in your economy, that's exactly what happens. Users, content, the network is

always going to prioritize a shorter path to getting to the same content that you

were trying to get. before. Ideally, you want that part to be a path that is

connected to the first Internet exchange point that is set up. Because if you have

more content, if you have more paths available to you via the Internet exchange

point, your users will use it. Contributes to, is lessening the outflow of capital

from your environment.

I should have prefaced this by saying that I'm an economist, I pretend to know BGP

and peering, and I pretended to know BGP and peering for 25 years. So, a lot of

what I think about, when it comes about peering, is talking about doing things at

national strategy level, because I see it as that way. I see it as reducing the

outflow of capital from your domestic economy.

The final part that we see here, and this isn't something I want to talk too much

about, is what happens when you get to a stage where your Internet service provider

networks have now migrated, or evolved, or grown to a point, where you're not just

carrying end users as a network, where the networks themselves, the customers of the

networks themselves, other networks, you become a network service provider in the

environment, you have multiple Internet exchange points connected to your city.

This is the end goal.

And I think that this is the thing that I am going to ask the panel about when we

get to the last part. So let's speak a little bit in detail about there. But I

always like to walk through. I like simple examples, because it helps me to think

about things. Let's walk through a simple concept that we think about, and one that

we use every single day, but might not necessarily be obvious to you in this way.

There's a concept called hot potato routing. Basically it describes how

packets move between networks. The concept of a hot potato, if you're trying to hold

it in your hand, is that you want to get rid of it as quickly as possible.

So, how do networks, and how does traffic move in this context of hot potato

routing? Well, we've got two networks again, network red and network green, and

we've got the red customer that's trying to send a packet to a green customer of

the network. In a situation where there are two Internet exchanges, how does

traffic flow here? Well, the red customer is trying to send traffic to the green

customer. So, we're moving from left to the right of the screen, the red customer

sends a packet to the red Internet service provider, its network, because the red

customer is paying the Internet service provider's network for transit, for

connection to the Internet. The red network has two choices, because it's

connected at two Internet exchanges, that's IXPS, and it can choose to either

send its traffic to Internet exchange 1 or Internet exchange 2.

There's a very simple concept that I want to perhaps sum all of this up by later.

But, for now, think of it as hot potato, how do I get rid of the hot potato as

quickly as I can? And what I'm going to do is, I'm going to pop that traffic off to

the Internet exchange that is closest to me. So, the red network makes its decision

to, say, let me push the traffic off to Internet exchange point 1. Because the red

and green networks are peering with each other, the green network sees the packet,

and because it does what responsible Internet service providers are meant to

do, it receives a packet from exchange point 1, it backhauls it across a longer

path. However long the path is not the point. What's important here is that

there's a short path and there's a long path. And the green network pulls the

packet back along the long path and delivers it to its customer. So far, we've

completed half of the transaction needed to support this Internet transaction. But,

as you know, most transactions on the Internet are always send and receive. So,

we've sent the packet, how does the return packet flow back?

Well, the return packet, we now have the green customer that's trying to speak to

the red customer. So, the green customer sends his packet into the green cloud, and

the green network has to make the same but reciprocal decision that the red network

made just a short while ago. How do I send this packet loss? And what it does, again,

simple concept, hot potato routing, it tries to get rid of the packet as quickly

as possible. And it sends it -- you might think it's logical, but we're going to

explain this, we're going to explore this in a second. But the green network hands a

packet off to the Internet exchange point 2. This time it's the job of the red

network to pull the packet across the long distance, and deliver the packet to its

customer. The end of the day, the customers are communicating with each

other, and the internet's working, and we're all happy.

But how does that work from a costing perspective, or here you've got two

Internet service sroviders connected to two Internet exchanges. Both of them would

have had to invest infrastructure, they would have to invest transport or

transmission to get connected to the Internet exchange. They may have to have

had to pay for routers that they have to install at each Internet exchange and, of

course, any other necessary infrastructure, personnel, etc, etc, to

build those connections. Because, in this very simple example, we can see that

because both operators are connecting to an exchange that's close to them, and one

that's far away from them, we have a relative sharing of costs. The red network

receives money from its customer, and it uses that money to invest in

infrastructure. Similarly, the green network receives money from the green

customers. It uses that to fund its bills to fund its connections to the exchanges.

And that's fantastic.

The short part of this is, we pretty much have symmetry, we have a fair sharing of

costs between the network operators. That's my point, when I said that peering

isn't really free, necessarily. You still have to pay to run and manage your

network. If you're a network service provider anyway, you have to pay to run

and manage your own network, you get investment from your customers, or

partners, or whomever, and you use that to build your network. We like to say at PCH

that the efficiency of the Internet depends on this principle, and that is, as

long as there are two exchanges, and as long as each partner is connected to both

Internet exchanges, and traffic is allowed to choose the path to flow across, and it

flows naturally in this way, the Internet kind of works out, everything becomes nice

and fair.

But, I prefer to think of the corollary here, I prefer to think about this from

the perspective of what happens if there are not two Internet exchanges. And for

me, the realization here is that countries, or cities, or economies that

haven't yet built Internet exchange points, well, you're disadvantaging

yourself, because all that you are doing if you do not have an Internet exchange

point in your local context, is that you are exporting capital outside of your

region. To put it in context, the little picture that we drew here, if you are the

green network, and there is no Internet exchange point that is close to you, you

are always going to be paying for the long distance circuit. And whether that long

distance circuit is to Frankfurt, Amsterdam, London, one of the more

interconnected regions of the world, you're paying the long distance cost. The

networks that you're connecting to on the other end are generally going to pay the

shorter costs. Always.

My challenge for you is for you to see this. This is what happens when you don't

have an exchange point in your economy, you're always paying the long distance

cost. How do we break out of this model? How do we break into a model where we can

get the networks, that we are sharing traffic with, to connect at multiple

Internet exchange points, or more importantly, connect to an exchange point

where we get to pay the short part of the circuit? As a very simple equation that

I'd love for you to remember, if there's nothing else you remember about this

presentation. remember two things: I didn't shave, I'm sorry, and secondly,

this simple equation, speed times distance equals cost. That's true when you build

and develop your network.

So, if this were a real face to face meeting, I would do a quick poll with you.

I would ask you, how many of your customers are interested in having less

speed? Which of your customers want slower Internet services? I've asked this

question all across the world, and there's only been one place -- I won't tell you

where, it's in the Caribbean, though -- somebody said, We'd like our customers to

have slower Internet service. But, across the world, it's generally considered that

people want more. That's it, people want more. They want more speed, and they want

it cheaper than what they're paying for now. So, you have two constraints in the

simple equation, you've got to increase speed, but you've got to decrease cost.

You don't need to be a master mathematician to realize that the only

variable that you are going to have to play with that's left here is distance,

which is what we were talking about. How do I reduce the distance? How do I reduce

the cost that I have to pay to get to the content or to get to the networks that I

need to exchange traffic?

So, I'm going to hand back to van and the rest of the panel. And I think these are

the some of the discussion points that you're going to work through. Vahan?

I'd like to thank the interpreter for the brilliant interpretation. He even

translates all the jokes. Anyway, we have a very interesting question about breaking

the vicious cycle. What do we do to build a domestic or original eco system? And how

do we make sure that we have a great local content? Because in the Soviet Central

Asia, we primarily have content which is generated elsewhere. How do we attract

global players to Central Asia? How do we minimize our connectivity costs? What's

your take on that? This is a question for all of the participants.

By the way, the very first question is probably the most relevant. Do Central

Asian countries have local content? Or do we go to mail.ru? Do we have any national

servers that could be developed to attract additional content providers to the

region?

If I may?

Sure, go ahead.

First and foremost, let me comment on the content. That's why I changed my

background. During the COVID-19 pandemic we realized that content provision is a

huge headache, especially as far as the educational content was concerned. In

Kyrgyzstan, we launched the Ilimbox project, which is aimed at developing

educational content in the local language because, indeed, we have training content

in Russian. If you speak English, you can enjoy much educational content in English.

However, Kyrgyz is a very popular language in rural areas, but there is very little

educational content in Kyrgyz language.

I changed my background on purpose to follow up on the issues raised by Nishal.

In his presentation he mentioned that it takes at least two IXPs to tango. Well, by

way of an example, ISPs and mobile carriers do not collaborate too much here.

Three mobile carriers build three data centers, or three ISP is deploy their own

cables. In other words, what I'm saying is that they do not cooperate with each

other, they want to build their own data centers, they want to deploy their own

cables, they want to build their own mobile communications towers, they use

separate power lines to power their data center, and so on and so forth.

Cooperation between the national players is a logical step.

On the face of it, we have to work for the same country, but in practice, they

compete with each other. We have to think about how to incentivize all the ISPs to

develop a joint IXP infrastructure. That's why a lot of our villages are not covered,

because in some regions, we have three towers, and in some regions, we don't have

a single tower. So, this is a problem. Thank you.

Thank you, Talant. And, by the way, this is a question which is linked with the

legal framework and regulations, the joint operation of telecom infrastructure and

Internet infrastructure, as well as the power grids, as well as other utilities. I

think that this issue is subject to different regulations in different

countries. We could address this question to Nishal and all the other panelists,

what's your take on that?

We will definitely raise this issue in September during the follow up event.

Because, like I said, regulations is one of the core issues for the Central Asian

countries. Access to the physical infrastructure is important, access to

utilities is also important. And you are right on point now that, during the

COVID-19 pandemic, access to both public resources such as educational resources,

and governmental body's resources became of paramount importance. I know for a fact

that many Central Asian countries passed laws, bylaws and regulations, obliging or

recommending to the carriers to ensure access to these resources. The same holds

true for the Caucasian countries and other regions.

So are there any other prerequisites to develop local networks? Oh, how do we

attract foreign participants? By the way, Nishal, are you talking about networks or

network carriers?

I see Michuki's hand is up. And I have a very good idea that we should he is going

to be answering that part of it as well.

Thank you, Vahan, and thank you, Nishal, for that great introduction and

background.

So, I think the conversation on this should be, one of the important things to

understand is the value that each network needs to bring -- will be bringing to the

Internet exchange point. And that's often an understated value in terms of why the

interconnection is important, or why is peering important. We've seen this, and

I'm speaking from a context of having worked with Nishal for the last decade or

so in Africa to build the peering and interconnection ecosystem. When people get

to an Internet exchange point and connect, they expect that by miracle, something,

that the traffic will just grow. And in actual sense, that's not the case. It's

understanding the role that you played, not just by building the infrastructure or

connecting the fiber to the ports on the switch, but the role that you actually

play as a community in first understanding how that values ecosystem works, or how

that value chain works. And second, the role that you will individually play to

bring this to grow that particular, or contribute to that growth. And this is

something that's often really understated in a lot of the conversations that we have

with stakeholders.

And so the question here is, how do we build the domestic content? So, we start

by understanding what is actually local, what services already exist locally that

are not available at the Internet exchange point. And we see many countries where

government services are not digitized. So, what is the role that the stakeholders can

play to influence government to actually digitise its services, and we've seen very

impressive statistics coming from some of these markets where previously government

services were not digitized. They've digitised over 200 or 300 of those

services. And that makes government become a content producing network, or CDN. And

that increases, not just the traffic, but the usage, because there is a demand now.

As a user, I need to access government services, I no longer need to go to an

office, I can access that electronically or digitally or via the Internet, and that

increases demand.

Secondly, the other services like banking services, which again have traditionally

been in person, but with mobile, the advancement of mobile services, etc, it

means you can have all services, these banking services, available digitally. And

that means that, if those services are available through the Internet exchange

point, then it drives users and increases the traffic that you will find locally.

And then we go now to the enterprise market, where you have a lot of other

enterprise businesses that are traditionally not seeing at the exchange

point, and haven't actually digitised their services. So, getting them to

digitise their services means more traffic available at the Internet exchange points

by virtue of networks, and so on.

The Research and Education Network is another network that is often not

connected to the Internet exchange point. In fact, some countries don't even have a

Research and Education Network. So, getting this one built to serve the

research and academia community, getting that connected to the Internet exchange

point, brings more traffic. And the national cctld, again, connected to the

Internet exchange point.

And so, you can now keep building on this. We've seen quite a lot of innovation

happening around this content space. We've seen betting services, which are

traditionally done using other means, and now the betting services connect to the

Internet exchange point, and so on. Yes, well, that might be controversial in some

countries, but we've seen them driving quite a lot of traffic, because if you

think of the last -- for those who are soccer fans who are watching the Euro 2021

matches, then you will see that draws a lot of betting behind which team will win.

And that, actually, really increases the amount of traffic. Or now, as it is,

watching those matches live and so on. So, getting the traditional media houses, TV

stations, etc. to connect, to digitise their services and connect online, will

suddenly be key to growing their networks, and also the resulting in the number of

content that will be available online.

As that grows, then it leads to the next question, which is how do you attract

foreign networks to participate. And this foreign networks means they're actually

going to be bringing two things. One, there will be either networks as initial

explained, that are actually looking to connect, they have customers in that

market. So, could be end users, could be enterprise. So, they could be either

content networks, or actually carrier networks that are looking to connect to

that market.

And, if the policy part, of course, needs to be in place, policy and regulation

needs to be in place, to make it easier for those foreign networks to actually

build and connect to those markets. Because remember, they're not coming to

sell services, they're just coming to peer. And we've seen countries that have

actually realized the value of this, and say, Oh, well, if they're only coming to

peer at the Internet exchange point, they don't need a license. So, they can come

in, put their infrastructure inside a data center, and immediately start peering

locally. That's a major milestone that will help attract foreign networks. Also,

making sure that the Internet exchange points operator actually makes it possible

for foreign networks to participate in its governance model, meaning that, Oh, can I,

as someone who's interested in IXPs from Kenya participate in the governance of the

Uzbekistan IXP. So, if that's the case, then that's great because it means there's

a lot more visibility into what's happening in the future, what's really

happening with the growth of the IXP, and as an interested party, then it means I

will then be pushing my organization to come and connect to that IX.

So, there is a lot more that will need to happen at the community level of the IXP.

And I think we'll talk about this a little bit more, about building that community,

strengthening the community role within the exchange point ecosystem, and

stakeholders, so that they can actually play a role in growing the networks and

traffic at the IXP. So I'll close there for now.

Thank you so much, Michuki. We have yet nother comment from Talgat on the

ontent. However, before I pass on the fl or to Talgat, I just wanted to raise anot

er issue. We're saying that the states sho ld allow IXPs and development. Even t

is phrasing means that we need to get per ission from the state authorities. But,

hat's just one part of the story. Anothe part is the governmental supp

rt. The government should ask the In ernet service providers to use th

ir local IXPs. This is especially importan for Central Asia. Do you think that gove

nmental support is important? Talgat? hat's your take on that?

Good afternoon, again, ladies and gentlemen. Actually, I wanted to offer a

couple of comments on the Central Asian content. It's not going to be in high

demand outside of Central Asia, let's be realistic, be realists. Even such European

countries as France or Germany don't even have websites or web portals that are in

demand globally. I would say there are a ew content providers, primarily the US

that covers the English speaking ustomer. China is a separate story be

ause it's a market within itself. And I be ieve that, in our region, Russia is the l

rgest content provider. By way of an e ample of the most popular content in K

zakhstan, we have a national singer na ed Erke Esmakhan, and she has over 130 m

llion views on YouTube for Central As a, this is quite a significant figure. An

we have other pop singers Kairat Nur as, Dimash, and they have 20 to 50 mill

on views on their YouTube channels. o, they have several dozen million v

ews on their YouTube channels. But, in C ntral Asia, we will never create anything

similar to YouTube. That's just an examp e, to illustrate what I mean.

As for the COVID-19 pandemic, and the issues around the online or e-learning, in

Kazakhstan, it was a huge problem and, I guess, this is the same situation for all

the other countries. Several million kids go to schools, but then during the

COVID-19 pandemic, for almost 45 days, all the high schools were shut down and they

had to study from home. All the services were based in the US, all the servers were

overloaded, the high schools gave up using zoom because that wasn't helpful. Teachers

would just send homework via email to students, and then they use some kind of

messenger to submit their homework to teachers. This situation happened just

because we did not have any local servers. We did not have any video conferencing

systems based on local service, and the traffic went to the backbone service.

In addition to IXPs, we need to develop local service and local content. The

COVID-19 pandemic will stay with us for years to come, so I guess all these issues

will remain relevant. In addition to creating IXPs, we need to create servers

such as Moodle, for instance, which is a popular server among the educational

communities, and they should be based in the clouds of such large carriers as

Kazakh Telecom. I'm talking about Kazakhstan only. That's in addition to

IXPs.

The last thing I wanted to mention, Talant, if my memory serves me right,

mentioned that, in Kyrgyzstan, one of the headaches is providing Internet access to

remote rural areas. This is a problem for all the countries even including the most

developed countries, but in 2020, in Kazakhstan, we completed a program called

250 Plus, which basically means that even the villages with a population of just

over 250 people had to be connected by cable to the Internet -- by the fiber

cable. And the mesh networks, mesh Wi Fi, were part of the program. ISOC, for some

reason, did not support this program. Sorry for that. Just wanted to mention

that in passing. Anyway, as far as the remote rural areas are concerned, I guess

the situation in Kazakhstan is better than that in other Central Asian countries. But

in a nutshell, this is it, brings me to the end of my remarks.

Thank you so much, Talgat. Before we move on to the next part, or the last part of

the presentation, I just wanted to pose a question to all the panelists. And

actually, I want to address this question to content providers, as well as people

who communicate with content providers on a daily basis. So, I always take Facebook

as an example. I was just wondering if there are any advantages for content

providers if they use IXPs, compared to telecom operators? For CDNs, what is more

preferable? IXPs or telecom operators? And Nishal, the floor is all yours.

And by the way, I'm sorry, for interrupting you. The reason I'm asking

this question is that the providers who have a certain volume of content. Facebook

and Google deploy their own CDN, and for them, it doesn't make any sense to peer

with IXPs. Nishal, the floor is yours.

Okay, thank you very much. I am going to politely disagree on that last point, I

happen to run three exchanges in my country. But, I think let's talk about

this, first of the slides, and we can come back to a discussion later.

So, 2019, the experience was that the Internet doubles in size almost every 11

months or so. Since 2020, when the Coronavirus has popped up, it's anybody's

guess as to how much faster it's grown since then. And the thing that you always

have to worry about, when you look at exponential growth like this, is one

question, and that is how do I scale this kind of growth? The way that we generally

would scale it is by saying that there are multiple elements that you have to worry

about when you build and design, not just the exchange point, but the elements that

fit in your ecosystem around that. You've already heard some of this come up

already, we've been speaking about Internet exchange points so far. But

exchange points are one critical element in this, what I call the cycle of

upgrades, right? There's other important things like local loops, it's pointless to

have an Internet exchange point, if network operators can't get to the

Internet exchange point, if the local loop is closed, or heavily restricted. If you

manage to do that, if you have high speed fiber, but there's high speed fiber only

within the confines of the metropolitan area, and you don't have any kind of

national backbone, it's really not going to help you if you have an Internet

exchange point.

Now, I'm not going to discuss each of these, firstly, because we're running low

on time. And also because their slides are here, so you can read through the slides,

and I think that the slides kind of go through the elements and explain the

different things that you need, the foundational building blocks of not just

the Internet exchange point, but the elements that are necessary to see the

economic growth that will continue to sustain the Internet exchange point but,

more importantly, the digital ecosystem that everybody's talking about building

inside their countries. I'm also led to believe that a lot of these issues, which

relate largely to regulation, how we should go about putting in place this kind

of infrastructure, that's really only going to happen, or those discussions are

going to happen in the next series of the Asia Peering Forum discussions, so I'll

leave those elements for you to talk about then.

But, for now, the takeaway I'd like you to have from me is this, that fixing the

exchange is not your silver bullet. It's certainly not going to be the element

that's going to say, we put in the exchange point, and what's going on?

Where's the growth that you promised us.? The exchange point is one part in what is

otherwise a very complex -- it's complex because there are multiple moving parts,

but it's really not that difficult to solve --system. Why I say that this is a

virtuous cycle is because you have to think about upgrading each of the

different components inside the cycle, and that upgrade cycle is usually a two year

cycle. So, in a two year span, we will talk about upgrading the exchange,

upgrading the local infrastructure, upgrading international submarine fiber

cables and critical infrastructure, you're going to get back to the point where

you're going to start all over again, it really is a continuous virtual cycle.

The complete picture looks more like this. There's a few more elements in here than

just the five that I had pulled out initially. When you sit down to discuss,

in your ecosystem, what are the things that we have? What are the things that we

should be building on? For me, this is a good starting block for you to work on.

You should be making a to do list, a checklist, that says, how do we better all

of these elements, and then continuously do that again? As I said, I won't discuss

this in too much detail, because this discussion is still happening at a little

later on.

What I'd like to talk a little bit about right now, though, is what it takes to

build and run the exchange. Because we're talking about developing Internet

exchanges, we're talking about getting exchange points operational, and there's

often confusion in how this should work. I pointed out, at the start, that I run

three exchange points in my country, I've been helping to build and develop Internet

exchange points for many years. Michuki and I have done a lot of good work

together, not just in Africa, but around the world. There are five key elements,

the five key elements that I've listed here, the ones that I think are really the

elements that you need to focus on and build, and that's really to consider.

Again, I'm not going to discuss all of this in detail, because I know the

panelists will. What is the governance structure that you want to think about for

running your exchange point? Here, my guidance to you is always that a community

built nonprofit Internet exchange point is almost always the best solution when

you're starting to think about building an Internet exchange. Unfortunately, we're

now living in a world where people think that it's okay to build an exchange for

profit. I don't like that idea, and I don't like the mentality that says the

exchange has to be profitable, because you shouldn't think of the exchange as

something that's making money. In fact, the exchanges that I promote, the

exchanges I try to get built, I usually encourage people to try to make them free

from cost. Because, in my experience, what happens is that if people start to see the

value of the exchange, then running into those elements that would otherwise cost

you money, things like, how do I get new equipment for the exchange? Things like,

how do I get technical support for the exchange? Or, how do I get dark fiber to

run, you know, multiple sites, if I'm doing that? That will come from your

environment. You're going to look at me and say, Nishal, you don't know what

you're talking about. The world is very different right here.

But, I'm thinking of 25 years ago, when the country that I live in, when

Johannesburg was a very different city, when we were asking ourselves those same

questions. How do we see this growing? How do we see this being built? How are we

going to get people involved? Because I want to say we got together as a

community, we recognize that from our position at the bottom of the world, if we

didn't try to make this work for ourselves, nobody else was going to build

infrastructure here. Because, over a period of 20 years, we've developed and

grown that environment, it's now I think that we are an attractive enough model for

big networks from the outside, that want to now come and invest and build

infrastructure and run infrastructure out of our country.

Remember what I said about the long path and the short path of the model, and it's

only really been in the last few years, that we've managed to get external

networks to come into South Africa. When that happens, when they build

infrastructure in your country, the cost to connected them becomes the cost of

getting fiber across town. Our community is now understanding that, they recognize

that, and that's why when I say community nonprofit, I can say that because that's

how we run our exchanges. Our community nonprofit donates time, they donate

people, they even donate dark fiber, we get dark fiber at no cost. Because folks

inside my country now understand that the bigger purpose of this Internet exchange

point, it's not just about lowering latency. It's not just about building

local content and growing local content. It's a larger economic picture. And if we

can do that together, in the form of a community, nonprofit, that becomes a lot

easier for people to want to invest in. And that's why my guide to you, when I

talk about governance is always to say, how do we make this work for us without

thinking of how do I necessarily make money out of it?

When I think about neutrality? Of course, neutrality is bestd, you will hear

fantastic stories about people saying that build carrier-neutral data centers, and a

carrier-neutral data center is the best for an exchange. I think we all know that.

But, the reality is that there isn't a large number of carrier-neutral data

centers in all over the world, or particularly in the area that you're

looking to build in. And, for me, what's more important is consensus. Consensus is

more important than cost, I write here. Don't build a carrier-neutral data center

if you want an Internet exchange point.

My go to story here is a country in West Africa that decided they needed an

exchange, so they took a loan from the World Bank for $2 million to build a data

center, so they could get an Internet exchange point. That is absolutely wrong

model for you to follow. Exchanges can develop anywhere, there are really big

exchanges that have started out in the broom closets of hotels. I can give you

lots of examples, but we're running out of time. What's important is that people that

are going to connect to the exchange, the networks that are going to connect, have

consensus in that where you're trying to put something, that's the most important

thing, that they are all happy with that location.

I've already spoken about the nonprofit. But for me really the part about being a

nonprofit is driving down the cost to operate. We have a saying at PCH that the

function of an Internet exchange is to lower the average per bit delivery cost of

the network, i.e. it's got to make the network cheaper to run. If you

artificially inflate the cost of running the Internet exchange point, if you start

to charge people extremely high port fees, if you start to hire people simply to run

the exchange, when you can have volunteers do this, then you've got to pay them

salaries upfront, even if you only have 10 peers. And it's really not a lot of work

to run a layer two fabric, right? If you do that you inflate the costs, the only

way you're going to get the cost back is by charging port fees, and that's going to

disincentivize people from connecting.

If you need equipment, there are lots of ISOC people on this call. ISOC, I believe,

donate equipment, PCH donates equipment, so you can write to me. But, there

shouldn't be a lot of friction to getting an exchange started once you have at least

three willing participants. Don't try, again, don't try to make money from the

Internet exchange.

I think the last point for me here really is this, the exchange must work to support

your community. This idea that we're going to build an Internet exchange, and

everybody is going to connect to it after we build it, is completely the wrong way

to think about it. The Field of Dreams approach -- Field of Dreams is a US movie,

if you're familiar with it, this idea that if we build it, it will work -- doesn't

work for Internet exchanges. What works for Internet exchanges is consensus within

the community. This has proven itself across the world, whether you're talking

about a large country that's perhaps sparsely populated, because people are all

over, or you're talking about highly densely built countries, island economies

as well, if you cannot get your community to have consensus that the exchange is

going to work and, in the domestic context, they're going to peer at the

Internet exchange, the exchange is not going to be successful, and that's the

wrong approach that you want to take to investing.

So, invest in your community, right? Invest in your community, because that's

the important thing. There's a line that I love to use, it's the line at the bottom

of the slide, and it says that, as operators, we all compete at commercial

level -- we have to because fundamentally, most of you are in this to make money,

right? So, we compete, obviously, at commercial level, but we collaborate a

technical level. We have to work with our competitors, even at least at a technical

level, to making this work, because it's that competitive collaboration that grows

the community, and that builds the Internet exchange for us to grow on.

There are a whole bunch of things that you can do at your exchange. I won't talk

about this, because we've already heard people mention this already. You can run

lots of anycast services, you can run validators, there's a series of services

that are available for you to do. I guess the most important thing here is, really,

how do I get the exchange to showcase that I have technical prowess in my country,

that the community is smart, and the community is engaged. We have a

competition really in my country, with the technical team, the volunteers that make

up the exchange, as to who can think of interesting services that we should have

at the Internet exchange, and usually the winner gets dinner, free dinner, and then

we all go and treat them to a dinner. That's the kind of competitive competition

that I think is useful, where you're actively thinking how do I make something

work better? That's the model I would definitely recommend for you.

Things that can go wrong? I've already spoken about increasing the cost. If you

don't want to increase the cost, don't look for work. Internet exchanges are

fundamentally very simple things to work. They are a layer two fabric as you start.

It's only when the exchange has to extend to multiple locations and have connections

across the city, and you should only really start doing that when exchanges

become -- well, rather, when the fiber to get from A to B becomes as close to zero

as possible. I mentioned that we get fiber donated to us, if we were not in a

position where we had fiber donated to us, then the cost of running darkt fiber to

different buildings, the cost of paying for colocation in all of those different

buildings, would be prohibitively expensive, would not allow us to run a

multi-site exchange. So, do what your community does, and our community has

asked us to run a multi-site exchange, and they have enabled us to do that through

things like donations from the data centers, give us free colocation, like a

cabinet where you can run exchange infrastructure, the fiber operators are

giving us fiber so that we can interconnect these things. When you get to

that point, that's your community voting for you. It's your community telling you,

we think you're doing the right thing, we want to be involved, we want to back that.

Don't try and run an old boys club. It's a common problem in a few areas of the world

where they think that the exchange point is really a connection place only for

Internet service providers. That was very true in probably the 90s, probably the

early part of the 00s, but it's no longer the case. If a network like Facebook, or

Google had to come to you and say, hey, we went to peer at your exchange, would you

tell them no? Of course not. You want the content at your exchange, right? So, make

the exchange inclusive, allow anybody that has a valid autonomous system number, and

valid Internet resources to come to the exchange and connect to the exchange.

You'll find that if you do that, people will choose whom they want to peer with,

and the exchange will grow.

That's the thing that you want, you want to build that ecosystem that the rest of

the world is going to look at, and say hmm, I'm going to invest there, because

that's how you attract foreign networks. That's your goal here. When foreign

networks get to your exchange, your cost to connect to them becomes close to zero

as possible. That's the long term picture we're working towards here.

And I don't bet on one horse. This is a bit of a weird thing for me to say but, I

always think of it as -- we have a saying at PCH, another saying, which is that

exchanges may develop differently, but we don't pick and choose which exchange is

the one that should develop. Maybe one exchange is going to run fundamentally

different to another. If you can, try and support them all. Because, ultimately,

it's going to be the community that really is going to decide what and how that

Internet exchange needs to operate, to work, to serve that local community.

I think for me, the telling part of all of this is, if you don't decide to do this,

if you don't decide to get up, and build, and grow your domestic exchanges in a

manner that works for you, then only two things are going to happen, only one of

two scenarios. You're going to continue to pay networks to ship your traffic

overseas, and you've heard some of the problems that brings, cost, time, latency

-- the example of changing traffic with the Chinese was really a good one, I

thought. You've heard about the problems that brings you, right? It's an export of

capital out of your economy and, obviously, that's not the situation you

want to be in. The only other viable option, if you don't decide to build and

grow your domestic exchange, is that somebody else will. You're going to find

probably a large exchange franchise going to come and try to deploy that in your

infrastructure, or somebody who thinks that they can build a large exchange

franchise. When they do it, when people that are not invested in your community

decide that they want to, how should I say, gift you an Internet exchange point,

they're not doing it for free, they're extracting capital from your economy,

again. So, either way, you're paying. And my challenge to you is act or pay. You

decide which model you want to build in.

I'm going to pause here, and I'm going to hand over to the participants, the

panelists, Vahan and the rest of the team for them to discuss the issues that we've

spoken about. I'm also just going to say, thank you. My name is Nishal Goburdhan, my

email address is on the screen if you'd like to contact me. I'd love to reach out

to some of you and hopefully be back in the region when the world turns to normal,

I hope. Take care and thank you.

Thank you so much, Nishal, for a very interesting presentation. Thank you for

giving us the step by step instructions on what to do, or how to do, and what not to

do.

The last but not least question pertains to the elements that are missing in the

local ecosystems. What do we do to ensure that we have good peering and great IXPs?

How do you identify these missing elements? And what do we do to deal with

those challenges? By the way, another important question is what IXP model is

best suited for a startup environment? Actually I will cross out startup, and I

will just say what IXP model is best suited for your countries. Is it a NPO

model, a commercial model, or a hybrid model? What do you think? Let me see if

there are any raised virtual hands. By the way, Aziz and Talant, you tried to use

this model in Kyrgyzstan, if I'm not mistaken. Go ahead, Talant.

in Kyrgyzstan, we used the open, non-commercial, nonprofit ISP, and all the

interested parties concerned can connect to this IXP. Of course, we're going to

have to look into the sustainability model. We'd like to thank ISOC for

supporting the construction of, the building of this global IXP. I liked what

Nishal stated, that we need to build consensus first, and then build the IXP.

Unfortunately, in our case, it was not really in that order.

Let me address Nishal's earlier question whether people want cheaper or slower

Internet connectivity. Nishal, let me share the following anecdote with you. In

Kyrgyzstan, we have the cheapest mobile Internet connectivity, we're only the

second best to Israel, according to the official statistics. But, on the other

hand, it's one of the slowest Internet connections in the world. So, in our case,

we choose to prefer the cheapest but the slowest Internet connection. Thank you so

much.

Aigerim, and then Stavros. Aigerim Abakirova, you have the floor.

Hi, everyone. My name is Aigerim Abakirova, and I work for the community of

carriers. Just for your information our association has built a an IXP that can

operate in both modes, namely both on a commercial basis, and on a nonprofit

basis. The nonprofit access is provided to members of our association, while the

commercial basis is offered to any other stakeholders who are willing to hook up to

our IXP. By the way, our IXP is one of the largest in Kyrgyzstan and, as far as I

know, there are no other IXPs of such a scale.

Our IXP is of a local nature, and to make sure that we get access to the global

backbone networks, last year as part of the [unintelligible] partnership program,

we pushed the following idea partnership with Deutsche-IX. The main reason for that

was that in Central Asia, each country has its own IXP. That holds for Kazakhstan,

Russia, and Kyrgyzstan. We started thinking, how come we don't have a

regional ISP that would bring together our national? And, actually, we already

offered two options to Deutsche IXP. The first one would be for them to build their

own IXP in one of the CAS member states, and we will all use them, or establish a

joint IXP in Kyrgyzstan, and [inaudible] IXP would become a regional IXP. This idea

was put on the back burner, because our authorities are still discussing these two

options. In a nutshell, this is it.

Thank you so much, Aigerim. Before I pass on the floor to Stavros, let me follow up

with another question. The two options that you described, do they include CDNs?

This is a question for the tech savvy guys, I'm a lawyer by background. I'm not

aware of that. I just described the situation as far as the regulations are

Michuki also raised his hand. So, you will have the floor after Stavros. Stavros, you

concerned.

have the floor, and then Michuki, you will hear the floor.

Thank you very much. Going back to your previous question, and coming back from

AMS-IX, which is a nonprofit organization, I believe this model of nonprofit and

neutral IXP is great, especially for economies that are currently growing, and

IXPs that are still in the very beginning of their life. That model is great

because, as history said, people will collaborate to solve some technical

issues, they work together to fix a lot of problems that they will exist initially.

As Nishal said in his presentation, with this great slide with the circle, it's not

only the IXP that has to be fixed, let's say, it's just one part, and is a very

simple part to fix. There are other things that need to be fixed as well. Colocation

is one, fiber connectivity is another one. And, of course, also the base with the

governance model and the regulations. So, all these parts need to be fixed. And this

part when they're going to fix, they're going to grow the IXP as well.

So, when people collaborate in a non-for-profit model, in a community based

model, they talk together, they try to solve the issues together. In the same

time they can push towards the government, or other parties, altogether in order to

bring their ideas into life. Coming back to AMS-IX history, that happened when

AMS-IX started up many years back, 25 years back, actually, when people had to

solve a simple technical problem, which was to share the same transatlantic link

that connects Europe and Netherlands with the United States. There was only one and

the problem that we're trying to fix was to -- how we share this common resource

among us, and this is how AMS-IX started, because people want to collaborate in an

efficient way to serve these resources. Then, later on, they worked together into

growing up the IXP. I believe this model is great. I don't say that the commercial

model is bad. Of course, it also has its own benefits. But, history says that,

okay, when people work in a collaborative and neutral way, the benefits are much

larger for a wide variation of the community as well. That was my comment in

the previous discussion and question.

Thank you Vahan, for the opportunity. I just want to comment on something that has

been mentioned by the speaker before Stavros, and this is with respect to the

issue of regional IXPS, or thinking that you can interconnect IXPS. It is something

that has come up in our region more than once, maybe even a couple of times, and it

has never been successful, and it is less likely to be successful for a number of

reasons.

First and foremost, let me start by stating that, when we talk about Internet

exchange points, there is not definitional understanding of a regional IXP. In fact

there is nowhere where you'll find an IXP stating itself to be regional. This is

because an IXP is there to serve networks that are operating in a particular market,

and you can have multiple IXPs in the same market. We've seen cities which have

multiple independently operated IXPs in the same city, in the same metropolitan

area, and so on. Part of that is because an IXP is serving the interests of its

community. So, it's not a national IXP. No, it's an Internet exchange point. So,

when people start referring to IXPs as national, regional, global, etc, then that

creates a certain level of misunderstanding, especially on the

policymakers and regulatory side of things, because that tries to insinuate

that also you can have a local IXP, which is connected to a regional IXP, which is

connected to the global IXP. But that's not exactly how that works, because the

way it works is that operator's networks connect at the Internet exchange point,

and Internet exchange points don't interconnect with each other. Because then

that will be going into the business of the operators. And that, as Nishal

mentioned, is that they will be now competing with the services of their

members, and that will mean members will then run away or disconnect from the

exchange point because it's now a competitor, rather than an enabler or

facilitator for them to get, you know, lower cost of access.

So that's really an important note thing to note. It is certainly a question that

pops up many times. You should not be surprised when it does. But we have had a

lot of lessons and discussions with policymakers with respect to this issue.

We're happy to talk to you maybe bilaterally, or on other sessions, on why

this is really important for you to raise up the understanding, with the

policymakers and regulators, of how exchange points work, and why we don't

interconnect them, and why the need really is for you to help support the

establishment of Internet exchange points to support the market where the demand is,

in other words, where there are more than three operators that are willing to

interconnect and there's consensus around that. Thank you.

That's a very interesting point of view. Thank you, Michuki, for sharing that.

Aigerim wants to take the floor.

I have a question for Michuki. He mentioned that right now there is no such

a thing as a regional IXP. I do understand that, it's just a definition that we use.

For instance, let's take Euro-IX, isn't it a regional IXP?

From the get go, be sure the answer is no.

Judging by the name, It's regional.

No, it's not. It's an association. Euro-IX, isn't it a regional iXP, that is

the question.

So, I've seen Bijal here, and probably would like to see her, because she comes

from Euro-IX, can talk about that. But maybe, just as she does, there are

regional associations that bring together exchange point operators, and provide them

with a platform where they can exchange and discuss and learn from each other. In

Africa, these are Af-IX, there's APIX in Asia, there is LAC-IX in Latin America. I

will invite Bijal to talk about the work that Euro-IX does so that there's better

clarity on that.

Yes. Thank you for the question. Just to confirm that Euro-IX is a membership based

Association for Internet exchange points. We're not an Internet exchange point.

We've never been an Internet exchange point. The idea is that, like Michuki

said, we're just a membership association. We have currently 71 members, and we come

together and work together to share information, learn. I hope that answers

the question.

Yes, thank you so much. Now it's clear. Since I'm quite familiar with the Euro-IX.

Let me add a few words to that. Euro-IX is not a virtual association, it does provide

multiple opportunities to IXPs to develop further. It assists IXPs in finding

partners. For instance, ARM-IX, an Armenian IXP received some hardware from

Euro-IX. AMS-IX donated some hardware to us as well. There are governance or

management systems, like IXP Manager, would be a good example of things that

Euro-IX is engaged in, and it also helps startup IXPs. Euro-IX is a down to earth

practical platform that can help the Central Asian region a lot in building a

good IXP model for it to take off.

I wanted to pose yet another question to all of the participants, which is as

follows: what issues would you like to discuss at follow up events, because the

reason we organize the series is for you to be able to discuss relevant issues for

your regional IXPs, or for the XPS that you want to develop in the region, for

them to be more sustainable, more robust. If you have any specific requests, I

understand that you might have multiple questions about the legal framework, the

regulations, we will discuss it at the next event. However, if you want to raise

additional issues, or if you want us to invite certain speakers or partners, just

let us know. The next event will take place in September, just let us know in

the chat box, or just raise your hand and state it orally. In addition to the

regulations and the legal framework, what other issues would you like to discuss

during the September event? Anything at all?

Vahan, hi.

Mavzuna, hi.

Hi, it's great to be able to hear you. First and foremost, I'd like to thank you

for organizing the series of events attended by experts from our countries.

You're well aware of the fact that we've been discussing this issue for quite a

number of years, but thanks to the COVID- 19 pandemic, the Internet boomed in our

countries. In the mega cities, broadband access has skyrocketed, and it's the

providers that initiated building an IXP. One of the questions is as follows.

Initially, we want to involve startups, young guys who want to take the lead, and

the very first question is, are there organizations or institutions that can

help us with some technical resources, administrative support, governance and

management skills. I believe that for countries that already have IXPs, it might

not be relevant, but for our countries, initially, some sort of support would be

important. So, I was just wondering if there are any institutions like that?

I think Mavzuna is based in Tajikistan. I guess the essence of the question is,

Tajikistan does not have its own IXP, if any of the panelists are ready take this

question now? At the last event, we discussed a very similar thing. Max, would

you like to comment on that?

Let me take that. I know that, time and again, that PCH provides, donates hardware

as well. Nishal is writing things in the chat box. PCH would be happy to assist,

you can email Nishal directly, he'll assist you on that. ISOC also donates

hardware, and it's not just network hardware. We can also provide the

auxiliary hardware such as solar panels, air conditioning units. The Euro-IX

participants can also provide some sort of assistance. The very first router that we

received was the one sent by AMS-IX. I'd like to thank both AMS-IX and the Euro-IX

for donating to us servers and network hardware, but the local community also

assisted us a lot. To get something from Europe or the US, first and foremost, it

takes time. And to build a small scale IXP you need to get some hardware from the

local carriers as well. RIPE NCC currently runs a program. Back then we did not have

a program like that, but now it's the projects fund.

The main message is that if there is a will, there is a way, and you will

definitely get assistance both from the global and the local stakeholders and

players. If you decide to build an IXP then all the counterparts including RIPE

NCC, ISOC, PCH, and other partners will come to the rescue. Trust me on that. You

can turn to any of us and we'll be more than happy to assist you. And if we're

unable to do that our partners will. Well thanks you so much. bijal@euro-ix.net. If

you have any questions pertaining to assistance, turn to one of the

counterparts.

I'd like to thank all of the participants. I'd like to thank all the panelists, and

I'd like to thank our sponsors for enabling us to organize a series of

events. I'd like to thank Facebook, FLEXOPTICS, AMS-IX, and I'd like to thank

in RIPE NCC.

Well, Hisham, thank you so much for organizing this event, and Nishal, thank

you for delivering the presentation. I would like to thank the interpreter for

the translation and for a sense of humor. Thank you so much.

If you have any other questions, you can email all of your questions to any of our

panelists. Thank you so much. I'll see you in September at the next event in our

series, you will get invitations in advance. Thank you so much.

The Description of Virtual Peering Series – Central Asia #2